Frequency band optimization Fourier decomposition method and its application in fault diagnosis of rolling bearings
HUANG Siqi
ZHANG Xinqun
LIU Shijie
TAN Zhiyin
HE Kai
Abstract：[Objective]The Fourier decomposition method(FDM)is a method that adaptively determines modal components based on signal spectral characteristics.However,when extracting modal components from non-stationary signals,this method tends to generate numerous invalid narrow-band components,which hinders the precise identification of fault features.To address this issue,a frequency band optimization Fourier decomposition method(FBO-FDM)was proposed.[Methods]Firstly,based on Fourier transform,the original Fourier spectrum was scanned and segmented in the order from high frequency to low frequency to obtain initial segmentation boundaries.Secondly,a frequency band reconstruction strategy was established.The partial mean of multi-scale permutation entropy(PMMPE)was used to quantify the frequency band information within each segmentation boundary,and bands with PMMPE values greater than the mean were retained to remove invalid narrow-band components.Finally,adaptive multi-scale morphological filtering was applied to the reconstructed components to eliminate the influence of noise and irrelevant components.The proposed method was analyzed using rolling bearing simulation signals and compared with FDM,empirical wavelet transform(EWT),and variational mode decomposition(VMD).[Results]The results show that FBO-FDM can more effectively identify fault characteristic frequencies with a higher signal-to-noise ratio(SNR),and exhibites better noise reduction performance for colored noise.When applied to the analysis of measured vibration signals,the comparative results further validate the superiority of FBO-FDM in frequency band division and fault diagnosis capability.
Keywords：Fourier decomposition methodPartial mean of multi-scale permutation entropyAdaptive multi-scale morphological filteringRolling bearingFault diagnosis
Publication Date：2025-09-15
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：11( 151-161 )